Holocene climate variability in the western Mediterranean

region from a deepwater sediment record

J. Frigola,1 A. Moreno,2 I. Cacho,1 M. Canals,1 F. J. Sierro,3 J. A. Flores,3 J. O. Grimalt,4

D. A. Hodell,5 and J. H. Curtis5

Received 21 April 2006; revised 3 December 2006; accepted 3 December 2006; published 4 May 2007.

[1] The detailed analysis of the International Marine Past Global Changes Study core MD99-2343 recoveredfrom a sediment drift at 2391 m water depth north of the island of Minorca illustrates the effects of climatevariability on thermohaline circulation in the western Mediterranean during the last 12 kyr. Geochemical ratiosassociated with terrigenous input resulted in the identification of four phases representing different climatic anddeepwater overturning conditions in the Western Mediterranean Basin during the Holocene. Superimposed onthe general trend, eight centennial- to millennial-scale abrupt events appear consistently in both grain size andgeochemical records, which supports the occurrence of episodes of deepwater overturning reinforcement in theWestern Mediterranean Basin. The observed periodicity for these abrupt events is in agreement with thepreviously defined Holocene cooling events of the North Atlantic region, thus supporting a strong Atlantic-Mediterranean climatic link at high-frequency time intervals during the last 12 kyr. The rapid response of theMediterranean thermohaline circulation to climate change in the North Atlantic stresses the importance ofatmospheric teleconnections in transferring climate variability from high latitudes to midlatitudes.

Citation: Frigola, J., A. Moreno, I. Cacho, M. Canals, F. J. Sierro, J. A. Flores, J. O. Grimalt, D. A. Hodell, and J. H. Curtis (2007),

Holocene climate variability in the western Mediterranean region from a deepwater sediment record, Paleoceanography, 22, PA2209,

doi:10.1029/2006PA001307.

1. Introduction

[2] The Holocene (last �10 kyr) has been classicallyconsidered a climatically stable episode, especially whencompared with climate changes of the last glacial period.However, there is increasing evidence of significant climatevariability at orbital and suborbital scales during the presentinterglacial [Bianchi and McCave, 1999; Bond et al., 2001;Magny et al., 2002; Kuhlmann et al., 2004; Mayewski et al.,2004; Alley and Agustsdottir, 2005].[3] Orbitally induced differences in seasonal insolation

have determined the long-term climatic evolution of theHolocene with a warm Climate Optimum during the early-to-mid Holocene and a transition to colder conditionsaround 5 ka [COHMAP Members, 1988; Cheddadi et al.,1997, 1998; Prentice et al., 1998; Claussen et al., 1999;Magny et al., 2002; Davis et al., 2003; Sbaffi et al., 2004].Superimposed on this pattern are events of rapid climatechange with periods of 2800–2000, 1500 and 900 years

[Mayewski et al., 2004]. While solar flux variability hasbeen proposed to be the main forcing of these Holoceneevents [O’Brien et al., 1995; Bond et al., 2001; Rohling etal., 2002; Mayewski et al., 2004], oscillations in theproduction rates of the North Atlantic Deep Water (NADW)and in the poleward heat transport could also have triggeredor amplified such instabilities [Bond et al., 1997; Bianchiand McCave, 1999; Schulz and Paul, 2002; Oppo et al.,2003]. In any case, the direct causative mechanism remainsunknown.[4] Paleoclimatic records have demonstrated the high

sensitivity of the western Mediterranean region to rapidclimate changes during the last glacial interval, includingDansgaard/Oeschger and Heinrich events, thereby support-ing the view of a strong link between the Mediterranean andthe North Atlantic climate [Rohling et al., 1998, 2002;Cacho et al., 1999, 2000, 2001; Moreno et al., 2002,2004; Martrat et al., 2004; Sierro et al., 2005]. This rapidconnection between both regions has been interpreted toresult from the entrance of cold surface waters into theMediterranean Sea through the Strait of Gibraltar, but alsofrom the intensification of the atmospheric circulation. Astrengthened westerly system enhanced the marine overturn-ing cell in the Gulf of Lion leading to a more efficientformation of Western Mediterranean Deep Water (WMDW)and to the enhancement of deep circulation [Cacho et al.,2001; Sierro et al., 2005].[5] In contrast to the glacial period, information about

Holocene rapid variability in the western Mediterraneanregion and its links to North Atlantic climate is compara-tively scarce. One of the most useful proxies for the study of

PALEOCEANOGRAPHY, VOL. 22, PA2209, doi:10.1029/2006PA001307, 2007ClickHere

for

FullArticle

1Consolidated Research Group Marine Geosciences, Department ofStratigraphy, Paleontology and Marine Geosciences, Faculty of Geology,University of Barcelona, Barcelona, Spain.

2Pyrenean Institute of Ecology, Spanish Research Scientific Council,Zaragoza, Spain.

3Department of Geology, University of Salamanca, Salamanca, Spain.4Department of Environmental Chemistry, Institute of Chemical and

Environmental Research-Spanish Research Scientific Council, Barcelona,Spain.

5Department of Geological Sciences, University of Florida, Gainesville,Florida, USA.

Copyright 2007 by the American Geophysical Union.0883-8305/07/2006PA001307$12.00

PA2209 1 of 16

WMDW formation and circulation during glacial periods,carbon and oxygen isotopic records from Cibicidoides sppforaminifera, is lacking during the Holocene because ofpoor ventilation and oxygenation conditions of deep watersthat caused the disappearance of this species [Caralp, 1988;Reguera, 2004]. Core MD99-2343 was recovered from thedeepwater Minorca sediment drift in the path of the south-ward branch of the WMDW. In this study, we use grain sizedistributions and bulk geochemical ratios of terrigenousmaterial down this core to reconstruct Holocene changesin WMDW.

2. Study Area

2.1. Climate and Physical Oceanography Setting

[6] The climate regime of the Mediterranean region istransitional between the temperate maritime type and thearid subtropical desert climate [Barry and Chorley, 1998].The Icelandic low–Azores high system controls present-daymeteorology and climate in western Europe including thewestern Mediterranean region. Summers in the westernMediterranean are usually hot and dry because of theinfluence of the expanded Azores anticyclone. The south-ward displacement of the anticyclone during winter allowsAtlantic depressions to enter the western Mediterraneanregion bringing high atmospheric instability and wetterconditions. At decadal scale, this pattern is known as theNorth Atlantic Oscillation (NAO), which modulates muchof the present-day climate variability in this region [Rodo etal., 1997]. The NAO system and the strong influence of theMediterranean Sea expose the region to large-scale climatechanges [Bolle, 2003]. The western Mediterranean region isalso influenced by Saharan air masses that transport con-siderable amounts of dust toward the Mediterranean Sea andfarther north [Prospero, 1996]. This short overview high-lights the complexity of the climatic behavior of the westernMediterranean region and evidences its high sensitivity toheat and moisture flux variations.[7] Because of winter southward displacement of the

Azores high, Atlantic depressions follow southern trajecto-ries coming into the Mediterranean region more frequently[Barry and Chorley, 1998]. This process leads to theformation of strong and cold northerly and northwesterlywinds in the Rhone and Ebro valleys funneling the airflowinto the western Mediterranean (i.e., Mistral and Cierzowinds, respectively). These winds cause strong evaporationand cooling offshore in the Gulf of Lion thus increasingsurface water density until it sinks to greater depths[MEDOC, 1970; Lacombe et al., 1985; Millot, 1999]. Thisprocess gives birth to the formation of the Western Medi-terranean Deep Water (WMDW), which fills the deepestpart of the Western Mediterranean Basin (Figure 1). Deepwater formation in the Gulf of Lion depends on wind stressvariability but is also affected by the amount and depth ofthe Levantine Intermediate Water (LIW) before WMDWformation events [Pinardi and Masetti, 2000]. As conse-quence of the negative balance of water created by theexcess of evaporation over fresh water input in the Medi-terranean Sea a compensating surface Atlantic water layerenters through the Strait of Gibraltar as Modified Atlantic

Water (MAW) (Figure 1) [Millot, 1999]. The dense LIW andWMDW leave the Mediterranean Basin through the Straitof Gibraltar forming the deep Mediterranean Outflow Water(MOW) [Millot, 1999].[8] In the northwestern Mediterranean Sea the Balearic

Promontory influences the circulation acting as a topo-graphic barrier. The dense WMDW that forms and sinksin the Gulf of Lion flows south and southwestward into theValencia Trough at depths closer to 2000 m [Millot, 1999](Figure 1). When the deep current encounters the BalearicPromontory it shifts direction eastward and southeastwardbordering the Minorca base of slope. The abruptness of theBalearic slope and the topographically induced change inthe current direction likely result in an intensification of thecurrent, as this process has been described for the NorthAtlantic deep sediment drifts [McCave and Tucholke, 1986].Off Minorca this has led to the formation of the Minorcaperipheral depression and associated sediment drift [Velascoet al., 1996] (Figure 2a) where our core MD99-2343 wasrecovered.

2.2. Particle Sources and Sedimentary Setting

[9] Sediment is supplied to the northwestern Mediterra-nean Sea mainly by fluvial discharge from the north, byaeolian inputs form the south, and by primary productionfrom surface waters. The two main rivers are the Rhone andthe Ebro (Figure 1) with estimated historical pre-dammingsediment fluxes of 30 � 106 t yr�1 and 17–25 � 106 t yr�1,respectively [United Nations Environment Programme,2003]. However, only 10% of the fluvial discharge reachesthe deep basin while the remaining 90% is deposited indeltaic and inner continental shelf areas [Martin et al.,1989]. Saharan dust fluxes account for 10–20% of present-day deep-sea sedimentation in the western Mediterranean[Loye-Pilot et al., 1986; Zuo et al., 1991; Guerzoni et al.,1997] although this contribution may have changed sub-stantially through time [Moreno et al., 2002; Weldeab etal., 2003]. The contribution of local pelagic, mostlycarbonate particles is limited by the oligotrophic characterof most of the western Mediterranean Sea [Bethoux et al.,1998]. In any case, at the location of the studied sedimentcore, carbonate may also have been contributed by shelfedge spillover processes from the nearby Balearic Prom-ontory [Maldonado and Stanley, 1979; Maldonado andCanals, 1982].[10] High-resolution seismic reflection profiles across the

Minorca drift show a reflector configuration that is typicalof contourite drifts (Figure 2b) [Vanney and Mougenot,1981; Stow, 1982; Stow et al., 2002]. While the peripheraldepression is filled with coarse sediment [Canals, 1980] it isassumed that the fine fraction escaped out of the depressionand contributed to the development of the sediment drift inits way toward the basin centre. The MD99-2343 site on theMinorca drift, and the drift itself, occupy a relativelyshallower position [Alonso et al., 1995] that is beyond thedirect influence of turbidite sedimentation (Figure 2a).However, it is likely that suspended particles escaping fromthe turbidite systems to the west (Ebro margin) and north(Gulf of Lion margin) may have been caught by the near-bottom circulation and added to the background sedimen-

PA2209 FRIGOLA ET AL.: HOLOCENE CLIMATE VARIABILITY

2 of 16

PA2209

tation of the Minorca drift (Figure 2a). Large-scale bedforms found in the deep northwestern Mediterranean Basinfurther indicate that bottom currents likely played a signif-icant role in the shaping of the seafloor, and thus insediment particle transport, winnowing and sorting in therecent past [Mauffret et al., 1982; Maldonado et al., 1985;Palanques et al., 1995; Acosta, 2005]. Although no currentmeter data exist for the vicinity of the core site, near-bottomcurrent measurements during a 3-month period at 1800 mwater depth in the Gulf of Lion deep margin, whereWMDW formation takes place, gave maximum values of50 cm s�1 and mean values of 20 cm s�1 [Millot andMonaco, 1984].

3. Material and Methods

[11] Sediment core MD99-2343 was recovered with aCalypso piston corer north of Minorca at 40�29.840N,04�01.690E and 2391 m of water depth in the northwesternMediterranean Sea (Figure 1), during Leg 5 of the R/VMarion Dufresne expedition within the International MarinePast Global Changes Study (IMAGES) programme. Fromthe total 32.44 m of core length, only the top 4 m

corresponding to the last 12 kyr are discussed in this paper.The top 4 m consists of grey nannofossil and foraminifersilty clay. Layers with high content of pteropod and gastro-pod shell fragments have been also observed all along theupper core section. As a general rule, one centimeter thicksediment samples were taken every 4 to 6 cm for oxygenand carbon isotope analyses of foraminifer shells, and grainsize and major element composition analyses of the bulksediment. Additional samples for grain size analyses werecollected at 2 cm resolution over selected intervals.[12] Samples for isotope analyses were washed over a

63-mm sieve and the retained fraction was dried and dry-sieved again using a 150-mm sieve. About 10 mg ofGloborotalia inflata and Globigerina bulloides werehand-picked for radiocarbon isotope analyses. The AMS14C analyses were performed in the U.S. National OceanSciencesAcceleratorMass Spectrometry Facility (NOSAMS).The ages were calibrated with the standard marine correc-tion of 408 years and the regional average marine reservoircorrection (DR) for the western Mediterranean Sea bymeans of the Calib 5.0.1 programme [Stuiver and Reimer,1993] and the MARINE04 calibration curve [Hughen et al.,2004].

Figure 1. Bathymetric map of the study area showing the general surface and deepwater circulationpatterns and the position of core MD99-2343. The box in the main map shows the location of Figure 2a,while the solid line illustrates the location of the seismic reflection profile in Figure 2b. WMDW isWestern Mediterranean Deep Water; MAW is Modified Atlantic Water.

PA2209 FRIGOLA ET AL.: HOLOCENE CLIMATE VARIABILITY

3 of 16

PA2209

[13] Approximately 5 to 10 specimens of Globigerinabulloides from the 300–350 mm size fraction were picked tomeasure stable isotope ratios. Foraminifer tests were soakedin 15% H2O2 to remove organic matter and sonicallycleaned in methanol to remove fine-grained particles. Theforaminifer calcite was loaded into individual reactionvessels and each sample was reacted with 3 drops ofH3PO4 (specific gravity = 1.92) using a Finnigan MAT

Kiel III carbonate preparation device. Isotope ratios weremeasured online using a Finnigan MAT 252 mass spec-trometer. Analytical precision was estimated to be ±0.08%for d18O and ±0.03% for d13C (1s) by measuring 8 stand-ards (NBS-19) with each carousel containing 38 samples.All isotope results are reported in standard delta notationrelative to V-PDB [Coplen, 1996].

Figure 2. (a) Detailed bathymetric map showing the main seafloor features nearby core MD99-2343.Shaded area roughly delimits the Minorca sediment drift. The abrupt step on the NE Minorca slope is theresult from the merging of the high-resolution swath bathymetry data set with the General BathymetricChart of the Oceans (GEBCO) digital database [Intergovernmental Oceanographic Commission et al.,2003]. (b) Very high resolution seismic reflection profile across the Minorca sediment drift and peripheraldepression (modified from Velasco et al. [1996]). The cross within a circle represents the direction of thecontour current that is normal to the image. Equivalent position of core MD99-2343 is also shown.

PA2209 FRIGOLA ET AL.: HOLOCENE CLIMATE VARIABILITY

4 of 16

PA2209

[14] Grain size was measured on the total fraction and thenoncarbonate fraction after removing organic matter andcarbonates by treatment with excess H2O2 and HCl, respec-tively. A Coulter LS 100 Laser Particle Size Analyser(CLS), which determines particle grain sizes between 0.4and 900 mm, was used to determine grain size distributionsas volume percentages. The laser diffraction size analyzerprinciple is based on the measurement of the diffractionangle produced by the particles when a laser beam goesthrough the sample in an aqueous solution. The correlationbetween diffraction angle and particle size is opposite[McCave et al., 1986]. Since diffraction is assumed to begiven by spherical particles, the resulting particle size is thatdiameter (known as equivalent spherical diameter). Subse-quently, laser diffraction methods are claimed to underesti-mate plate-shaped clay mineral percentages. To correct sucheffect we have followed the method proposed by Konertand Vandenberghe [1997]. CLS precision and accuracy istested by systematic control runs using latex microsphereswith predefined diameters. The high precision (reproduc-ibility) of the measurements was demonstrated by smallvariations in the mean diameter (0.97% of variation) and inthe standard deviation (1.37% of variation). The accuracy ofthe measurements, as indicated by the relative departurefrom the nominal mean diameter is 0.30%, corresponding toabsolute deviations between 0.09 and 0.34 mm. Additionaltest runs were performed using microsphere assemblageswith mixed grain sizes to ensure that CLS accuratelydetermined polymodal grain size distributions.[15] We discuss grain size results as the median of each

sample since it represents the distribution midpoint and itusually constitutes a more representative value of the grainsize distribution than the mean. In order to extract palae-oclimate information from a mixture of sediments withdifferent sources numerical-statistical modeling of largegrain size data sets provides the best results [Weltje andPrins, 2003]. However, core MD99-2343 was recovered ona contouritic drift built by the influence of near bottomcurrents where minor or no changes in sediment sources are

expected (see below). Subsequently, instead of statisticalmodeling of end-members, we have considered the UP10fraction, which composes the volume percentage of thefraction coarser than 10 mm, a good indicator of deepcurrents variability at this site. The UP10 integrates thesortable silt fraction (SS, 10–63 mm), defined as the coarserfraction of the silt with noncohesive behavior during trans-port and deposition [McCave et al., 1995], while taking alsointo account the influence of the fine sand subpopulation(>63 mm) that could be reworked by strong contour cur-rents. Finally, the silt/clay ratio has also proven to be usefulfor the study of deepwater currents intensity [Hall andMcCave, 2000].[16] The percentages of major elements in sediment

samples were determined by means of X-ray fluorescenceusing a Philips PW 2400 sequential wavelength X-rayspectrometer. Prior to the analyses, samples were dried at100�C, and then ground and homogenized in an agatemortar. 0.3 g of homogenized bulk sediment with lithiumtetraborate at a 1:20 dilution factor were fused at 1150�C inan induction oven Perle’X-2 to 30-mm-diameter glass discs.The content of major elements Si, Ti, Al, K, Ca, Fe, Mn,Mg, P and Na was calculated as oxide percentages. Ana-lytical accuracy was checked by measuring internationalstandards (GSS-1 to GSS-7) being better than 1% ofcertified values. Precision of individual measurements wasbetter than 0.9% as determined from replicate analyses ofsediment samples (repeatability). Precision over the periodof measurement was better than 3.4% (reproducibility) forall elements analyzed in this work. Spurious correlationsbetween elements due to closure effect to 100% are avoidedby discussion of element/Al ratios [Rollinson, 1993].[17] X-ray diffraction (XRD) analyses were carried out in

selected samples using a Siemens D-500 X-ray diffractom-eter on untreated, glycolated and heated (550�C) samples.Prior to the analysis, samples were mounted on smear slidesafter clay separation by decantation.

4. Chronostratigraphy

[18] Sierro et al. [2005] provided an age model for coreMD99-2343 based on four 14C AMS dates and several tiepoints with the Greenland ice core GISP2. This age modelhas been modified to account for six additional monospe-cific foraminifer 14C AMS dates of which four lie within the0–12 ka interval (Table 1). A mid Termination Ib (TIb)additional tie point has been added by correlating theMinorca G. bulloides oxygen isotopic record to the onefrom the Alboran Sea core MD95-2043 [Cacho et al., 1999](Table 1). Both oxygen isotopic profiles for the last 12 kyrare plotted in Figure 3.[19] Sedimentation rates for the upper 4 m of core MD99-

2343 range between 18 and 73 cm kyr�1 with an averagevalue of 37 cm kyr�1 (Figure 3). These rates are much higherthan those determined for nearby sites (e.g., 4 cm kyr�1 incore SL87 south of Minorca [Weldeab et al., 2003]), evenaccounting for potential core stretching [Skinner andMcCave, 2003]. The high rates likely reflect local en-hancement of particle deposition associated with the build-ing of the Minorca sediment drift, as illustrated by seismic

Table 1. Age Model for Core MD99-2343a

Isotope Event or RadiocarbonSample/Foram Type

Depth,cm

14C Age,years

CalendarYears

AMS 14C/multispecific 28 790 (±40) 386 ± 55AMS 14C/G. inflatab 88 3,110 (±30) 2,816 ± 50AMS 14C/multispecific 118 3,390 (±50) 3,225 ± 80AMS 14C/G. inflatab 208 5,720 (±40) 6,091 ± 70AMS 14C/multispecific 238 6,210 (±50) 6,601 ± 70AMS 14C/G. inflatab 308 7,700 (±40) 8,110 ± 60T1b, onset of the Holocene 354 10,696c

AMS 14C/G. bulloidesb 398 10,650 (±50) 11,883 ± 230AMS 14C/G. inflatab 418 11,200 (±50) 12,811 ± 30AMS 14C/G. bulloidesb 568 13,850 (±40) 15,912 ± 190AMS 14C/multispecific 604 14,550 (±110) 16,822 ± 240

aNew and previous 14C AMS dates after Sierro et al. [2005] calibratedwith the Calib 5.0.1 programme [Stuiver and Reimer, 1993]. Linearinterpolation between dated points was performed with the AnalySeriesVersion 1.1 [Paillard et al., 1996].

bNew 14C AMS dates.cTie point used for the age model of core MD99-2343 by correlation with

the oxygen isotopic record from core MD95-2043 in the near Alboran Sea.

PA2209 FRIGOLA ET AL.: HOLOCENE CLIMATE VARIABILITY

5 of 16

PA2209

reflection profiles (Figure 2b). Because of the variation insedimentation rates as a consequence of the sedimentaryenvironment and because of the number of available dates,the final age model was constructed by linear interpolationbetween calibrated ages instead of using other age-depthextrapolation models [Telford et al., 2004]. The age modelaccuracy and the sampling interval result in a mean timeresolution of 135 years for the top 4 m section of coreMD99-2343.

5. Results

5.1. Oxygen Isotopic Record

[20] The heaviest values in the G. bulloides d18O recordfrom core MD99-2343 during the last 12 kyr correspond tothe late Younger Dryas (12–11.5 ka) (Figure 4). During thedeglaciation (11.5–9 ka), the d18O record shows a contin-uously decreasing trend that ends at 9 ka when the lightestd18O values were reached. The Holocene is characterized bya long-term rising trend punctuated by nine centennial tomillennial-scale oscillations (Figure 4a). Some of the d18Oincreases are significant, e.g., >0.5% from 6.5 to 5.8 ka, or>0.9% from 9 to 7.8 ka. Moreover, as discussed below,these oxygen isotopic anomalies (arrows in Figures 4, 5,and 6 and Table 2) correlated with changes in other proxies(see below). We name these events as ‘‘Minorca abruptevents’’ with M8 being the oldest and M0 the youngest. M0is not very well expressed in the d18O record but we have

also labeled this event considering the other studied proxies.The duration and intensity of these d18O shifts are similar tothose recorded during some of the glacial period Dansgaard/Oeschger cycles [Sierro et al., 2005] and only M0 may fitwithin the d18O analytical error. Following previous results[Cacho et al., 1999; Shackleton et al., 2000; Skinner andShackleton, 2003; Sierro et al., 2005] these d18O fluctua-tions would be driven by SST coolings of about 2� to 3�C.However, the influence of other properties (i.e., salinity) onthe isotopic signal cannot be discarded with the availableinformation.

5.2. Grain Size Distribution

[21] Down-core trends in median grain size are similar forbulk sediment and the noncarbonate fraction (Figures 4band 4c). The median grain size ranges between 5 and 8 mm,thus pointing to the same processes controlling the deposi-tion of the two fractions. Only a small number of samplesfrom the noncarbonate fraction have median grain sizescoarser than 10 mm. Since the carbonate fraction integratesbiological production plus detrital carbonate particles, it isconsidered that the fraction better representing the intensityof bottom currents is the noncarbonate one [McCave et al.,1995]. This noncarbonate fraction displays seven mediangrain size peaks at 8.4, 7.2, 6.2, 5, 4.1, 3.2 and 2.5 ka, whichare coincident with isotopic enrichment events, i.e., M8 toM2 (Figures 4a and 4c).

Figure 3. Comparison of the G. bulloides oxygen isotopic records from (a) MD95-2043 (Alboran Sea)and (b) MD99-2343 (this study) cores for the last 12 kyr. (c) Sedimentation rates along MD99-2343sediment core calculated linearly among calendar years from 14C accelerator mass spectrometry (AMS)dates (triangles) and tie points (cross) utilized in the age model (see text for details and Table 1). Themean sedimentation rate of 37 cm kyr�1 is represented by a dashed line.

PA2209 FRIGOLA ET AL.: HOLOCENE CLIMATE VARIABILITY

6 of 16

PA2209

Figure 4. (a) G. bulloides oxygen isotopic record from core MD99-2343 for the last 12 kyr. Grain sizerecords of the (b) total and (c) noncarbonate fraction of the sediment, expressed as the median (mm) ofeach sample. The image is a plane light photograph of the noncarbonate fraction coarser than 63 mm from304-cm core depth. Asterisks indicate bimodal samples. (d) Examples of unimodal (solid line, samplebetween M5 and M6) and bimodal (dotted line, from one M7 sample) grain size distributions. (e) UP10fraction (>10 mm) and (f) silt/clay ratio for the noncarbonate fraction (solid line) and the total fraction(dashed line). Arrows and bars indicate the position of the Minorca abrupt events M8 to M0 as defined inthe text. The 14C AMS dates (triangles) and tie point (cross) are also shown.

PA2209 FRIGOLA ET AL.: HOLOCENE CLIMATE VARIABILITY

7 of 16

PA2209

Figure 5. (a) G. bulloides oxygen isotopic record from core MD99-2343 for the last 12 kyr. (b) Si, (c) Ti,(e) K, and (f) Ca geochemical records from core MD99-2343 normalized to Al. (d) Summer insolationcurve at 40�N for the last 12 kyr. A dashed line between Si/Al and Ti/Al ratios represents the four distinctphases described in the general trend (see text). Arrows and grey bars indicate the Minorca abrupt eventsM8 to M0 as defined in the text. The 14C AMS dates (inverted triangles) and tie point (cross) are alsoshown.

PA2209 FRIGOLA ET AL.: HOLOCENE CLIMATE VARIABILITY

8 of 16

PA2209

Figure 6. (a) Continuous and (b) 300-year running mean records of the oxygen isotopic profile from theGISP2 ice core [Grootes et al., 1993; Meese et al., 1997]. (c) G. bulloides oxygen isotopic recordfrom core MD99-2343 for the last 12 kyr. (d) UP10 fraction (>10 mm) record for the noncarbonatefraction. (e) Si/Al profile with an indication of the four phases identified (I–IV). The arrows and the greybars represent the nine Minorca abrupt events M8 to M0. Dashed lines represent an attempt to correlatethe Minorca events with the oxygen isotopic record from the GISP2 ice core. The 14C AMS dates(inverted triangles) and tie point (cross) are also shown.

PA2209 FRIGOLA ET AL.: HOLOCENE CLIMATE VARIABILITY

9 of 16

PA2209

[22] A detailed study of grain size distributions from thenoncarbonate fraction reveals that most of the samples withinthe Holocene are unimodal, with mode values around 8–10 mm (Figure 4d). Few samples show bimodal distributionswith a second mode around 150–200 mm (Figure 4d), whichis given by the presence of coarser grains, mainly quartzand mica packets, as observed by microscope (see photo-graph in Figure 4c). Such bimodal samples (marked withan asterisk in Figure 4c) correspond, with rare exceptions,to the samples with high median grain size values in thenoncarbonate fraction and hence to the Minorca abruptevents M8 to M2.[23] The UP10 general trend mimics the median grain size

records and highlights additional features (Figures 4b, 4c,and 4e), for example a marked decrease from 11.5 to 10 ka,and a better expression of M4 (Figure 4a). From 9 ka topresent, the UP10 fraction oscillates between 30 and 35%,except for M8 to M2 events. In the M8 to M2 peaks theUP10 fraction may represent as much as 50% of thesample (Figure 4e). The UP10 peaks are consistent withthe above mentioned median grain size increases (Figure 4c),and correlative with the oxygen isotopic enrichments of theMinorca abrupt events (Figure 4a). A slight increase in theUP10 fraction is recognized in the last millennium.[24] Silt/clay ratio (Figure 4f) complements the informa-

tion gathered from grain size parameters. Like UP10, thisratio also illustrates a marked fall of silt accumulationduring the deglaciation, displays relatively high values dur-ing the early Holocene and shows a decreasing trend duringthe last 9 kyr ending with relatively low values during the lateHolocene (Figure 4f). The overall record is again punctuatedby a number of peaks linked to the Minorca abrupt eventsidentified from previous proxies. These long-term trends andshort-lived events are visible in both the bulk (dashed linesin Figure 4) and the noncarbonate fraction (solid lines inFigure 4), where they are more obvious.

5.3. Geochemical Record

[25] The Si/Al, Ti/Al, K/Al and Ca/Al geochemicalprofiles of core MD99-2343 are plotted in Figures 5b,5c, 5e, and 5f. Four main phases or trends are identifiedthrough the last 12 kyr: I, a decreasing trend in Si/Al and

Ti/Al ratios (Figures 5b and 5c) during the deglaciationthat leads to a minimum at 10.5 ka that coincides withminimum values in grain size proxies (Figure 4) andmaximum values of summer solar insolation at 40�N(Figure 5d); II, an increasing trend in both Si/Al and Ti/Alratios coincides with the end of the second phase ofTermination (TIb) and the early Holocene (10.5–7 ka); III,high Si/Al and Ti/Al ratios with moderate oscillations duringthe mid-Holocene (7–4 ka); and IV, a gradual decreasingtrend in Si/Al and Ti/Al during the late Holocene (4–0 ka),which parallels the decrease in the insolation curve(Figure 5d). Interestingly, the K/Al record (Figure 5e)presents a distinctive pattern during TIb and the earlyHolocene (phases I and II), therefore suggesting the operationof differentiated controlling factors on K/Al. The Ca/Al ratio(Figure 5f) shows a rather distinct pattern with an almostcontinuous increasing trend across the Younger Dryas, TIband early and mid-Holocene, with maximum values between4 and 2.4 ka that drop abruptly after 2.4 ka (Figure 5f).[26] The above described general trends are punctuated by

oscillations lasting from centuries to millennia. The Si/Alrecord shows eight abrupt events during the Holocene thatare centered at 8.4, 7.2, 6.2, 5, 4.1, 3.2, 1.6 and 0.5 ka(Figure 5b). Most of these events can be identified in theTi/Al and Ca/Al records as well (Figures 5c and 5f).However, they are not noticeable in the K/Al record(Figure 5e). The abrupt events in the geochemical recordscoincide with the Minorca abrupt events identified in theG. bulloides oxygen isotopic record and the grain sizeproxies. The only exception to the described overallpattern is M2, which is represented by one of the largestisotopic excursions (Figure 5a), and a Ca/Al peak butlacks of expression in the Si/Al record (Figures 5f and 5b).[27] The main mineralogical components obtained from

the clay size XRD analysis confirm the high percentage ofcalcite and illite within all the analyzed samples, withchlorite, kaolinite, quartz and very low percentages offeldspars. The lack of major changes in the mineralogicalcomposition of the clay fraction shows that geochemicalvariability is dominated by the composition of the coarserfraction. This enhances the value of the geochemical

Table 2. Timing of Holocene Abrupt Climate Eventsa

EventCentralAge, ka

Time SincePreviousEvent, ka

AgeInterval,

kaDuration,

ka

Cold Events in the North Atlantic and Mediterranean RegionsGlobal

Compilationof Events

SSTAlboran

Lakes andRivers Mediterranean

IRDNorth Atlantic

Salt and DustGreenland

M0 0.5 1.1 0.8–0.2 0.6 - 0.8 - 0.6–0 0.6–0.15M1 1.6 0.9 1.8–1.4 0.4 1.4 2 1.4 - 1.2–1M2 2.5 0.7 2.6–2.3 0.3 - - - 3.1–2.4 –M3 3.2 0.9 3.4–3.1 0.3 - 3 2.8 3.5–2.5M4 4.1 0.9 4.2–4 0.2 - 4 4.2 - 4.2–3.8M5 5 1.2 5.3–4.7 0.6 5.4 - - 6.1–5 6–5M6 6.2 0.9 6.5–5.8 0.7 - - 5.9 - -M7 7.2 1.5 7.4–6.9 0.5 - 7 - - -M8 8.4 - 9–7.8 1.2 8.24 9 8.1 8.8–7.8 9–8

aTimings of the Holocene Minorca abrupt events found in core MD99-2343 and tentative correlation with abrupt events recorded from SST in theAlboran Sea [Cacho et al., 2001], in lakes and rivers from the Mediterranean region [Magny et al., 2002], in ice-rafted detritus (IRD) from the NorthAtlantic region [Bond et al., 1997], in salt and dust from Greenland ice [O’Brien et al., 1995], and the compilation of Holocene rapid climate change eventsfrom Mayewski et al. [2004].

PA2209 FRIGOLA ET AL.: HOLOCENE CLIMATE VARIABILITY

10 of 16

PA2209

records for the interpretation of processes controllingcoarse particle release, transport and accumulation.

6. Discussion

6.1. Particle Sources

[28] Sediment particles resulting from riverine influx,aeolian transport and sea surface biogenic production (seesection 2.2) form a mixed population that is expected tosettle in the water column where it is advected by watermass movements before final deposition on the seabed. Theproxies used in our study allow identifying the ultimatefactors controlling sediment deposition in the MD99-2343core site.[29] The location of the core MD99-2343 on a sediment

drift off the carbonate shelf of the Balearic Islands points toa mixed signal in our Ca/Al ratio resulting from carbonateproductivity [Ruhlemann et al., 1999] and resedimentedcarbonate particles [Van Os et al., 1994]. Surface produc-tivity and subsequent particle settling in such an oligotro-phic area [Bethoux et al., 1998] can contribute only partly tothe carbonate flux and to the relatively high sedimentationrates of core MD99-2343.[30] Most of the samples within the Minorca abrupt

event layers display a characteristic bimodal distribution(Figures 4c and 4d) that could be tentatively attributed topulses of enhanced aeolian transport. The relatively lowrates of Saharan dust deposition and the high sedimentationrates measured in our core lead us to consider the aeoliancontribution as largely diluted within particle populationsfrom other sources. In addition, the coarser grains fromthese layers yield a 150–200 mm mode that is much coarserthan the one found in modern and glacial Saharan dustsamples in the western Mediterranean [Guerzoni et al.,1997; Moreno et al., 2002]. Furthermore, microscopeinspection of the coarse grains observed within theMinorca event layers shows quartz grains with moderateangularity and undisturbed mica packets (sea photographin Figure 4c), an uncommon feature in aeolian dustparticles [Guerzoni et al., 1997]. Those observations pointto a rather proximal source for these coarse grains whoserelease and transport did not involve particularly aggressivephysical or chemical weathering processes as would be thecase for the aeolian transported particles.[31] Core MD99-2343 was recovered from a contourite

drift and the occurrence of this coarse grain population maybe related to the formation of the drift itself. The building ofthe contourite drift demonstrates the efficiency of deepwatercirculation in the area to rework, winnow, transport andaccumulate originally fluvial terrigenous particles from theValencia Valley and therefore explains the relatively highsedimentation rates observed in core MD99-2343. Weinterpret changes in the grain size distribution as mostlygoverned by the strength of such deep currents as previ-ously observed in other contourite systems [McCave andTucholke, 1986; Llave et al., 2006; Voelker et al., 2006].Intervals of enhanced currents resulted in a more efficienttransport of coarse particles from both far fluvial sourcesand local sources as pointed out by the presence of coarsequartz and mica grains with minimal alteration. A promi-

nent volcanic ridge [Maillard and Mauffret, 1999] at thevery head of the Minorca peripheral depression (Figure 2a),to the west of the Minorca drift, is a firm candidate assource area for such unaltered particles. Therefore thecoarse particles deposited during the Minorca abrupt eventsaccumulated during intervals of near-bottom currentstrengthening able to vigorously erode seafloor relievesand transport to the sediment drift location the coarseparticles thus released.[32] K/Al, Si/Al and Ti/Al ratios are associated with

terrigenous inputs [Krom et al., 1999; Wehausen andBrumsack, 1999, 2000; Moreno et al., 2001, 2002;Martınez-Ruiz et al., 2003; Weldeab et al., 2003; Morenoet al., 2005], probably from the Ebro and Rhone riversnorth of the study site. Si mostly comes from alumino-silicates and quartz as biogenic opal is a minor sedimentcomponent in the region [Weldeab et al., 2003], as furtherconfirmed by our microscopic examinations. Ti resideswithin heavy minerals such as ilmenite and rutile, and Aland K are associated with clay minerals. In particular, theK/Al ratio is considered a good indicator for clay inputs(mainly illite) from river runoff. The presence of notice-able amounts of illite at the study site has been confirmedby peaks in XRD diffractograms. Consequently, the K/Alratio can be interpreted as an indicator of illite entranceby river discharge and hence may provide a diagnosis ofhumidity conditions in the northwestern Mediterraneanregion. The parallelism between the K/Al record andthe insolation curve at 40�N not only for the Holocene(Figures 5d and 5e) but also for the last glacial period(J. Frigola et al., Evidences of abrupt changes inWestern Mediterranean Deep Water circulation duringthe last 50 kyr: A high-resolution marine record fromthe Balearic Sea, submitted to Quaternary International,2006) reinforce the view that precipitation controls long-term K/Al ratio oscillations. The differences observedbetween Si/Al and Ti/Al, and the K/Al record (Figures 5b,5c, and 5e) are attributed to grain size geochemical segrega-tion processes since K is mostly associated with clay particleswhile Si and Ti relate to coarser grains. Parallel increases ofthe median grain size and the Si/Al and Ti/Al ratios supportthe view that grain size distribution controls the variability ofthese geochemical ratios rather than changes in source area.

6.2. Holocene Onset and General Trends

[33] The decrease in the oxygen isotopic record from 12 to9 ka (Figure 4a) embraces the end of the Younger Dryas, thesecond phase of Termination (TIb), and the onset of theHolocene. The variations observed in both grain size andgeochemical records during this time interval (Figures 4 and 5)reflect the strong changes in the sedimentary dynamics drivenby the shifting climatic conditions. Diminutions in grain sizeparameters and Si/Al and Ti/Al ratios (phase I), which reachedminimum values between 10 and 11 ka (Figures 4 and 5), areconsistent with a slowdown of deepwater circulation in thewestern Mediterranean. Accordingly, the glacial benthic iso-topic record from our core ends at 12 kawhenC. pachydermusdisappears from the benthic assemblage [Reguera, 2004;Sierro et al., 2005], likely replaced by species inhabitingpoorly oxygenated environments [Caralp, 1988; Reguera,

PA2209 FRIGOLA ET AL.: HOLOCENE CLIMATE VARIABILITY

11 of 16

PA2209

2004]. Reduced deepwater ventilation conditions for this timeinterval (12–9 ka) are also suggested by the preservation of anOrganic Rich Layer in the Alboran Sea [Cacho et al., 2002]. Inparallel, TIb sedimentation rates were minimal (Figure 3c)because of the combined effect of the reduction in the input ofterrigenous particles forced by the inshore migration of thecoastline caused by the postglacial sea level rise, and toa lowered transport of particles into weakened deepwatercurrents.[34] The weakening of the deep overturning cell could

also result (or be amplified) from more humid conditions inthe western Mediterranean region during the time of max-imum summer insolation (Figure 5d) as suggested frommaximum values of the K/Al record (Figure 5e) andsupported by other studies [Harrison and Digerfeldt,1993; Gonzalez-Samperiz et al., 2006]. A more pronouncedstratification in the upper water column favored by anenhanced freshwater input due to increased precipitationduring summer insolation maxima, and higher atmosphericstability associated with the retreat of the Northern Hemi-sphere ice sheet, would have lead to the slowdown of thedeepwater overturning cell in the Gulf of Lion and to areduction of contour current activity in the Minorca sedimentdrift. After 10.5 ka both grain size and geochemical recordsshow a steady increasing trend (phase II in Figures 5 and 6),which points to the recovery of the deepwater overturningcell in the Gulf of Lion coincident with the decreasing trendin the summer insolation at 40�N (Figure 4d).[35] The relative stabilization of grain size parameters

(Figures 4e and 4f) and the Si/Al ratio around 7 ka(phase III in Figure 6e) is synchronous with the end of thepostglacial sea level rise [Fleming et al., 1998] and suggeststhe reestablishment of deepwater circulation and a stablesupply of fluvial material. Such a synchronicity illustrateshow significant was the sea level rise control on the westernMediterranean thermohaline circulation and, consequently,on the outbuilding of the Minorca sediment drift. Theessentially stable conditions found for the mid-Holocene(7–4 ka) in the Minorca deep sea site contrast with markedchanges reported in many locations worldwide [Steig, 1999]and, in particular, in the Mediterranean borderlands[COHMAP Members, 1988; Cheddadi et al., 1997; Prenticeet al., 1998;Magny et al., 2002] and the North African region[Vernet and Faure, 2000] associated with the end of theAfrican Humid Period [deMenocal et al., 2000]. This mid-Holocene climate variability is attributed to the reduction ofseasonal insolation differences after 5.5 ka, which lead to anabrupt transition from humid to arid conditions in NorthAfrica and in the western Mediterranean region.[36] This well-known mid-Holocene variability does not

seem to influence our proxies until 4 ka when an evidentdecrease in the silt/clay ratio and the Si/Al points to aslowdown of the deepwater overturning cell and to a reduc-tion of fluvial inputs due to drier conditions (Figure 4f andphase IV in Figure 6e). These drier conditions would be alsoconsistent with the d18O stabilization at high values duringthe late Holocene. A southward displacement of the ITCZand the subsequent decrease in the atmospheric pressuregradient due to reduced seasonal insolation differences likelyfavored the establishment of drier conditions [McDermott

et al., 1999; Jalut et al., 2000]. Accordingly, the lessening inthe activity of the northwesterlies would account for thereduction of deepwater circulation during the late Holocene.Our results stress the high sensitivity of the western Medi-terranean thermohaline circulation to both the atmospheric(i.e., northwesterlies variability that induced changes in thedeepwater overturning in the Gulf of Lion) and the hydro-logic systems (i.e., orbitally induced precipitation variabilityand meltwater pulses).

6.3. Holocene Abrupt Events

[37] The nine Holocene d18O enrichment events had anaverage duration of 500 years and an observed periodicityclose to 1000 years (Table 2 and Figure 6). Most of theHolocene increases in the oxygen isotopic record parallelincreases in the UP10 fraction and the Si/Al ratio (Figure 6),therefore suggesting that relatively cold surface conditionscoexisted with more energetic deepwater conditions. Anintensification of the northwesterly winds in the westernMediterranean would account for the conditions describedduring the Holocene Minorca events, similarly to themechanism proposed for the glacial Dansgaard-Oeschgervariability [Cacho et al., 2000; Sierro et al., 2005]. Thesecold and dry winds enhanced the deepwater overturning inthe Gulf of Lion by cooling of the surface waters and,consequently, they steered the activity of bottom currents onthe Minorca rise. Such vigorous currents were able totransport coarser particles to the Minorca rise as shown bythe accumulation of quartz grains and mica packets coarserthan 63 mm. This resulted in the increase of the UP10fraction (Figure 6d) and the apparition of bimodal samples(Figure 4c). In parallel, higher values in the silt/clay ratioare interpreted as resulting from the winnowing effect of thefinest particles (Figure 4f).[38] While events M8 to M3 are consistently represented

in both grain size and geochemical proxies, M0, M1 and M2are not always well represented. For instance, M2 is notrecorded by the Si/Al ratio while it forms one of the largerpeaks in the UP10 fraction and, in contrast, M1 and M0 donot show a clear expression in the UP10 ratio but theypresent significant increases in the Si/Al record (Figure 6).Interestingly, this distinctive sensitivity between the differ-ent proxies occurs during the phase IV, late Holocene,related to the establishment of drier conditions due toreduced seasonal insolation differences (section 6.2). Over-all, reduced deep overturning is interpreted to occur becauseof the weakening of northwesterlies and more stable con-ditions. Furthermore, the expression of the Minorca eventsin the studied proxies seems to become weaker through theHolocene in parallel to the relative stabilization of theoxygen isotopic signal. These different climatic boundaryconditions may have determined a lower sensitivity of thesystem to millennial-centennial-scale climatic variabilitytoward late Holocene and, consequently, provided an am-biguous signature in the studied records.[39] The fact that Holocene cooling events have been

reported elsewhere all around the globe [Mayewski et al.,2004] demonstrates their global extent and the lack ofstability of the Holocene climate. There are, however,disagreements about the precise timing, the character and

PA2209 FRIGOLA ET AL.: HOLOCENE CLIMATE VARIABILITY

12 of 16

PA2209

the impact of these Holocene abrupt events. Some of thecooling events recorded in the Atlantic and Mediterraneanregions are summarized and tentatively correlated to ourMinorca cold events M0 to M8 for the last 9 kyr (Table 2).Furthermore, a correlation attempt between the Minorcaevents and the 300-year running mean of the GISP2 isotopiccurve (Figure 6b) results in fairly good agreement, evenconsidering the uncertainties of our age model. This sup-ports the hypothesis of a highly efficient climatic couplingbetween the North Atlantic and the western Mediterraneanregion during the last 10 kyr.[40] The UP10 fraction of our core MD99-2343 displays a

periodic oscillation close to 900 years in between 9 and 2 ka(Figure 6e and Table 2), which is in agreement with thoseobtained from the GISP2 ice core oxygen isotopic record[Schulz and Paul, 2002], from a Saharan dust record[Kuhlmann et al., 2004], and from varved sediments inCalifornia [Nederbragt and Thurow, 2005]. Several hypoth-eses aim at explaining the periodicity of about 900 years.Though it may be triggered by insolation changes or resultfrom internal feed back mechanisms [Schulz and Paul,2002], the recorded climatic oscillations are better linkedto the temperature signal from the North Atlantic climatesystem, in agreement with other records from the NorthernHemisphere [Cacho et al., 2001; Kuhlmann et al., 2004].[41] The most pronounced Holocene abrupt event, M8,

occurred at 9–7.8 ka, therefore embracing the well-known8.2 ka cold North Atlantic event [Mayewski et al., 2004;Alley and Agustsdottir, 2005; Rohling and Palike, 2005].The intensification of the atmospheric circulation during M8led to good ventilation conditions in the Western Mediter-ranean Basin thus stopping the formation of the ORL in theAlboran Sea [Cacho et al., 2002], synchronously with themiddle interruption of sapropel S1 in the Eastern Mediter-ranean Basin [Rohling et al., 1997; Mercone et al., 2000].We propose that the atmospheric teleconnection betweenhigh latitudes and the Mediterranean region through thewesterly winds system was the main control over thewestern Mediterranean thermohaline circulation throughthe Holocene. This atmospheric forcing of the climatevariability for the last 10 kyr is quite similar to the presentpattern of the North Atlantic Oscillation (NAO) that exerts afirst-order control at decadal scales [Rodo et al., 1997].Positive NAO years are associated with Iberian dryness andcold temperatures in Greenland, and more persistent andstronger winter storms crossing the Atlantic Ocean [Hurrell,1995]. Consequently, the Minorca abrupt events could beassociated with periods of persistent positive NAO index,which would strengthen the northwesterlies over the north-western Mediterranean Basin and hence reinforce deepwateroverturning. Although we do not exclude a pervasive solarinfluence or instabilities inherent to the North Atlanticthermohaline circulation [Bond et al., 2001; Schulz andPaul, 2002] as main precursors of the Holocene climateoscillations recorded in the core MD99-2343, we suggestthat the rapid transmission of these changes from highlatitudes to the Mediterranean region was mainly drivenby the northwesterly wind system variability modulated bya NAO-like mechanism. A similar atmospheric linkagemechanism, though acting at a millennial scale, was pro-

posed to explain the Dansgaard/Oeschger variability in thedeepwater ventilation and Saharan dust input during the lastglacial period in the Alboran Sea [Cacho et al., 2000;Moreno et al., 2002; Sanchez-Goni et al., 2002].

7. Conclusions

[42] The sedimentary record from core MD99-2343recovered from a deepwater contourite drift reveals theeffects of Holocene climate variability over the thermoha-line circulation in the Western Mediterranean Basin duringthe last 12 kyr. Geochemical proxies associated withterrigenous inputs like Si/Al, Ti/Al and K/Al display adecreasing trend through the Holocene that parallels thesummer insolation curve at 40�N showing the markedinfluence of the precipitation pattern over the region. Fourdifferent phases have been identified in the Si/Al and Ti/Alratios from the last 12 kyr. The first from 12 to 10.5 kashows a slowdown of deepwater circulation due to thecombined effect of the increasing sea level and therelatively humid conditions installed on land which bothfavored the stratification of water masses. The secondphase (10.5–7 ka) is associated with the recovery ofdeepwater circulation until the end of the postglacial sealevel rise at 7 ka. The third phase (7–4 ka) corresponds toa plateau with high values of the terrigenous proxiestranslating the good functioning of deepwater circulationduring a progressive orbitally driven change toward dryerconditions in the westernMediterranean borderlands. Finally,the fourth phase (4–0 ka) indicates a progressive decrease ofthe terrigenous contributions because of reduced fluvialinputs during drier conditions induced by lower seasonalinsolation differences, also modulated by the thermohalinecirculation weakening because of more stable atmosphericconditions.[43] Superimposed on this general Holocene pattern,

marked oscillations have been noticed and related to abruptclimate changes. The new grain size parameter presented inthis work that represents the fraction coarser than 10 mm(UP10) has been tested as a convenient proxy for paleo-current intensity in the study area for Holocene sediments.The UP10 record presents a 900-year cycle oscillation,which is consistent with the geochemical record of terrig-enous input between 9 and 2 ka and the surface coolingevents uncovered by the oxygen isotopic record. Suchperiodicity fits with temperature oscillations from the Holo-cene d18O record in Greenland and points to the pressuregradient system as a direct teleconnection mechanism forclimate variability transfer from the North Atlantic to theMediterranean region. The centennial to millennial-scaleHolocene oscillations observed in our records reveal acoupled atmospheric/oceanographic forcing equivalent tothe present-day NAO and sustains the hypothesis of a rapidfitting with Mediterranean climate conditions. Furthermore,our results demonstrate the high sensitivity of deepwateroverturning in the Gulf of Lion to the transfer of climateoscillations from high latitudes to midlatitudes.

[44] Acknowledgments. Funding by the European Commission Fifthand Sixth Framework Programmes to projects ADIOS (EVK3-2000-00035), PROMESS 1 (EVR1-CT-2002-40024), EUROSTRATAFORM

PA2209 FRIGOLA ET AL.: HOLOCENE CLIMATE VARIABILITY

13 of 16

PA2209

(EVK3-2002-00079), and HERMES (GOCE-CT-2005-511234-1) sup-ported the research effort behind this paper. The Spanish-fundedBTE2002-04670 and REN2003-08642-C02-02 projects are equally ac-knowledged. We are especially grateful to the Marion Dufresne and theIMAGES programme that enabled the collection of cores MD99-2343 andMD95-2043. GRC Geociencies Marines is recognized within the Generalitatde Catalunya excellence research groups program (reference 2005SGR00152). We thank M. Guart (Departament d’Estratigrafia, Paleontologia i

Geociencies Marines, University of Barcelona) and E. Seguı (ServeisCientifico-Tecnics, University of Barcelona) for their helpwith the laboratorywork andG. Lastras andD.Amblas for their helpwith the artwork. The EditorG. Dickens, J. B. Stuut, and two anonymous reviewers are greatly acknowl-edged for their positive comments on an earlier version of the manuscript.COMER Foundation and I3P postdoctoral programme (CSIC) are alsoacknowledged for their support to I. Cacho and A. Moreno, respectively.J. Frigola benefited from a fellowship of the University of Barcelona.

ReferencesAcosta, J. (2005), El Promontorio Balear: Morfo-logıa submarina y recubrimiento sedimentario,Ph.D. thesis, 154 pp., Univ. de Barcelona e Inst.Espanol de Oceanogr., Barcelona.

Alley, R. B., and A. M. Agustsdottir (2005), The8k event: Cause and consequences of a majorHolocene abrupt climate change, Quat. Sci.Rev., 24, 1123–1149.

Alonso, B., M. Canals, A. Palanques, and J. P.Rehault (1995), A deep-sea channel in thenorthwestern Mediterranean Sea: Morphologyand seismic structure of the Valencia channeland its surroundings, Mar. Geophys. Res., 17,469–484.

Barry, R. G., and R. J. Chorley (1998), Atmo-sphere, Weather and Climate, 7th ed., 409 pp.,Routledge, London.

Bethoux, J. P., P. Morin, C. Chaumery, O. Connan,B. Gentili, andD. Ruiz-Pino (1998), Nutrients inthe Mediterranean Sea, mass balance and statis-tical analysis of concentrations with respect toenvironmental change, Mar. Chem., 63, 155–169.

Bianchi, G. G., and I. N. McCave (1999), Holo-cene periodicity in North Atlantic climate anddeep-ocean flow south of Iceland, Nature, 397,515–517.

Bolle, H.-J. (2003), Mediterranean Climate:Variability and Trends, 372 pp., Springer,New York.

Bond, G., W. Showers, M. Cheseby, R. Lotti,P. Almasi, P. de Menocal, P. Priore, H. Cullen,I. Hajdas, and G. Bonani (1997), A pervasivemillennial-scale cycle in North Atlantic Holo-cene and glacial climates, Science, 278, 1257–1266.

Bond, G., B. Kromer, J. Beer, R. Muscheler,M. Evans, W. Showers, S. Hoffmann, R. Lotti-Bond, I. Hajdas, and G. Bonani (2001), Persis-tent solar influence on North Atlantic climateduring the Holocene, Science, 294, 2130–2136.

Cacho, I., J. O. Grimalt, C. Pelejero, M. Canals,F. J. Sierro, J. A. Flores, and N. J. Shackleton(1999), Dansgaard-Oeschger and Heinrichevent imprints in Alboran Sea temperatures,Paleoceanography, 14, 698–705.

Cacho, I., J. O. Grimalt, F. J. Sierro, N. J.Shackleton, and M. Canals (2000), Evidencefor enhanced Mediterranean thermohaline cir-culation during rapid climatic coolings, EarthPlanet. Sci. Lett., 183, 417–429.

Cacho, I., J. O. Grimalt, M. Canals, L. Sbaffi,N. Shackleton, J. Schonfeld, and R. Zahn(2001), Variability of the western Mediterra-nean Sea surface temperature during the last25,000 years and its connection with the North-ern Hemisphere climatic changes, Paleoceano-graphy, 16, 40–52.

Cacho, I., J. O. Grimalt, and M. Canals (2002),Response of the western Mediterranean Sea torapid climate variability during the last 50,000years: A molecular biomarker approach,J. Mar. Syst., 33-34, 253–272.

Canals, M. (1980), Sedimentos y procesos en elmargen continental sur-Balear: Control clima-tico y oceanografico sobre su distribucion yevolucion durante el Cuaternario superior,M. S. thesis, 210 pp., Inst. ‘‘Jaume Almera’’(CSIC) y Univ. de Barcelona, Barcelona.

Caralp, M. H. (1988), Late glacial to recentdeep-sea benthic foraminifera from the north-eastern Atlantic (Cadiz Gulf) and westernMediterranean (Alboran Sea): Paleoceano-graphic results, Mar. Micropaleontol., 13,265–289.

Cheddadi, R., G. Yu, J. Guiot, S. P. Harrison, andI. C. Prentice (1997), The climate of Europe6000 years ago, Clim. Dyn., 13, 1–9.

Cheddadi, R., H. F. Lamb, J. Guiot, and S. vander Kaars (1998), Holocene climatic change inMorocco: A quantitative reconstruction frompollen data, Clim. Dyn., 14, 883–890.

Claussen, M., C. Kubatzki, V. Browkin, andA. Ganopolski (1999), Simulation of an abruptchange in Saharan vegetation in the mid-Holocene,Geophys. Res. Lett., 26, 2037–2040.

COHMAP Members, (1988), Climatic changesof the last 18,000 years: Observations andmodel simulations, Science, 241, 1043–1052.

Coplen, T. B. (1996), New guidelines for thereporting of stable hydrogen, carbon, and oxy-gen isotope ratio data, Geochim. Cosmochim.Acta, 60, 3359–3360.

Davis, B. A. S., S. Brewer, A. C. Stevenson, andJ. Guiot (2003), The temperature of Europeduring the Holocene reconstructed from pollendata, Quat. Sci. Rev., 22, 1701–1716.

deMenocal, P., J. Ortiz, T. Guilderson, J. Adkins,M. Sarnthein, L. Baker, and M. Yarusinsky(2000), Abrupt onset and termination of theAfrican Humid Period: Rapid climate re-sponses to gradual insolation forcing, Quat.Sci. Rev., 19, 347–361.

Fleming, K., P. Johnston, D. Zwartz, Y.Yokoyama,K. Lambeck, and J. Chappell (1998), Refiningthe eustatic sea-level curve since the Last GlacialMaximumusing far- and intermediate-field sites,Earth Planet. Sci. Lett., 163, 327–342.

Gonzalez-Samperiz, P., B. L. Valero-Garces,A. Moreno, G. Jalut, J. M. Garcıa-Ruız,C. Martı-Bono, A. Delgado-Huertas, A. Navas,T. Otto, and J. J. Dedoubat (2006), Climatevariability in the Spanish Pyrenees duringthe last 30,000 yr revealed by the El Portaletsequence, Quat. Res., 66, 38–52.

Grootes, P., M. Stuiver, J. W. C. White, S. J.Johnsen, and J. Jouzel (1993), Comparisonof oxygen isotope records from the GISP2and GRIP Greenland ice cores, Nature, 366,552–554.

Guerzoni, S., E. Molinaroli, and R. Chester(1997), Saharan dust inputs to the westernMediterranean Sea: Depositional patterns,geochemistry and sedimentological implica-tions, Deep Sea Res., Part II, 44, 631–654.

Hall, I. R., and I. N. McCave (2000), Palaeo-current reconstruction, sediment and thorium

focussing on the Iberian margin over the last140 ka, Earth Planet. Sci. Lett., 178, 151–164.

Harrison, S. P., and G. Digerfeldt (1993),European lakes as palaeohydrological and pa-laeoclimatic indicators, Quat. Sci. Rev., 12,233–248.

Hughen, K., et al. (2004), Marine04 marineradiocarbon age calibration, 26 – 0 ka BP,Radiocarbon, 46, 1059–1086.

Hurrell, J. W. (1995), Decadal trends in theNorth Atlantic Oscillation: Regional tempera-tures and precipitation, Science, 269, 676–679.

Intergovernmental Oceanographic Commission,International Hydrographic Organization, andBritish Oceanographic Data Centre (2003)Centenary Edition of the GEBCO Digital Atlas[CD-ROM], Br. Oceanogr. Data Cent., Liver-pool, U.K.

Jalut, G., A. Esteban Amat, L. Bonnet, T. Gauque-lin, and M. Fontugne (2000), Holocene climaticchanges in the western Mediterranean, fromsouth-east France to south-east Spain, Palaeo-geogr. Palaeoclimatol., Palaeoecol., 160,255–290.

Konert, M., and J. Vandenberghe (1997), Com-parison of laser grain size analysis with pipetteand sieve analysis: A solution for the under-estimation of the clay fraction, Sedimentology,44, 523–535.

Krom, M. D., A. Michard, R. A. Cliff, andK. Strohle (1999), Sources of sediment tothe Ionian Sea and western Levantine basinof the eastern Mediterranean during S-1 sapro-pel times, Mar. Geol., 160, 45–61.

Kuhlmann, H., H. Meggers, T. Freudenthal, andG. Wefer (2004), The transition of the mon-soonal and the N Atlantic climate system offNWAfrica during the Holocene, Geophys. Res.Lett., 31, L22204, doi:10.1029/2004GL021267.

Lacombe, H., P. Tchernia, and L. Gamberoni(1985), Variable bottom water in the WesternMediterranean Basin, Prog. Oceanogr., 14,319–338.

Llave, E., J. Schonfeld, F. J. Hernandez-Molina,T. Mulder, L. Somoza, V. Diaz del Rio, andI. Sanchez-Almazo (2006), High-resolutionstratigraphy of the Mediterranean outflowcontourite system in the Gulf of Cadiz duringthe late Pleistocene: The impact of Heinrichevents, Mar. Geol., 227, 241–262.

Loye-Pilot, M. D., J. D. Martin, and J. Moreli(1986), Influence of Saharan dust on the rainacidity and atmospheric input to the Mediter-ranean, Nature, 321, 427–428.

Magny, M., C. Miramont, and O. Sivan (2002),Assessment of the impact of climate andanthropogenic factors on Holocene Mediterra-nean vegetation in Europe on the basis ofpalaeohydrological records, Palaeogeogr.Palaeoclimatol. Palaeoecol., 186, 47–59.

Maillard, A., and A. Mauffret (1999), Crustalstructure and riftogenesis of the Valencia

PA2209 FRIGOLA ET AL.: HOLOCENE CLIMATE VARIABILITY

14 of 16

PA2209

Trough (north-western Mediterranean Sea),Basin Res., 11, 357–379.

Maldonado, A., and M. Canals (1982), El mar-gen continental sur-balear: Un modelo depos-icional reciente sobre un margen de tipopasivo, Acta Geol. Hisp., 17, 241–254.

Maldonado, A., and D. J. Stanley (1979), Deposi-tional patterns and late Quaternary evolution oftwo Mediterranean submarine fans: A compar-ison, Mar. Geol., 31, 215–250.

Maldonado,A.,H.Got,A.Monaco,S.O’Connell,and L. Mirabile (1985), Valencia Fan (north-western Mediterranean): Distal deposition fanvariant,Mar. Geol., 62, 295–319.

Martin, J.-M., F. Elbaz-Poulichet, C. Guieu,M.-D. Loye-Pilot, and G. Han (1989), Riverversus atmospheric input of material to theMediterranean Sea: An overview, Mar.Chem., 28, 159–182.

Martınez-Ruiz, F., A. Paytan, M. Kastner, J. M.Gonzalez-Donoso, D. Linares , S. M.Bernasconi, and F. J. Jimenez-Espejo (2003),A comparative study of the geochemical andmineralogical characteristics of the S1 sapro-pel in the western and eastern Mediterranean,Palaeogeogr. Palaeoclimatol. Palaeoecol.,190, 23–37.

Martrat, B., J. O. Grimalt, C. Lopez-Martinez,I. Cacho, F. J. Sierro, J. A. Flores, R. Zahn,M. Canals, J. H. Curtis, and D. A. Hodell(2004), Abrupt temperature changes in thewestern Mediterranean over the past250,000 years, Science, 306, 1762–1765.

Mauffret, A., M. Labarbarie, and L. Montadert(1982), Les affleurements de series sedimen-taires pre-pliocenes dans le bassin Mediter-raneen nord-occidental., Mar. Geol., 45,159–175.

Mayewski, P. A., E. E. Rohling, J. C. Stager,W. Karlen, K. A. Maasch, L. D. Meeker, E. A.Meyerson, F. Gasse, S. van Kreveld, andK. Holmgren (2004), Holocene climate varia-bility, Quat. Res., 62, 243–255.

McCave, I. N., and B. E. Tucholke (1986), Deepcurrent-controlled sedimentation in the wes-tern North Atlantic, in The Geology of NorthAmerica, Volume M, The Western North Atlan-tic Region, edited by P. R. Vogt and B. E.Tucholke, pp. 451–468, Geol. Soc. of Am.,Boulder, Colo.

McCave, I. N., R. J. Bryant, H. F. Cook, andC. A. Coughanowr (1986), Evaluation of alaser-diffraction-size analyzer for use withnatural sediments, J. Sediment. Res., 56,561–564.

McCave, I. N., B.Manighetti, and S. G. Robinson(1995), Sortable silt and fine sediment size/composition slicing: Parameters for paleocur-rent speed and paleoceanography, Paleocea-nography, 10, 593–610.

McDermott, F., S. Frisia, Y. Huang, A. Longinelli,B. Spiro, T. H. E. Heaton, C. J. Hawkesworth,A. Borsato, E. Keppens, and I. J. Fairchild(1999), Holocene climate variability in Europe:Evidence from d18O, textural and extension-rate variations in three speleothems, Quat.Sci. Rev., 18, 1021–1038.

Medoc, G. (1970), Observation of formation ofDeep Water in the Mediterranean Sea, 1969,Nature, 227, 1037–1040.

Meese, D. A., A. J. Gow, R. B. Alley, P. M.Zielinski, G. A. Grootes, M. Ram, K. C. Tay-lor, P. A. Mayewski, and J. F. Bolzan (1997),The Greenland Ice Sheet Project 2 depth-agescale: Methods and results, J. Geophys. Res.,102, 26,411–26,423.

Mercone, D., J. Thomson, and I. W. Croudace(2000), Duration of S1, the most recent sapro-

pel in the eastern Mediterranean Sea, as indi-cated by accelerator mass spectrometryradiocarbon and geochemical evidence,Paleoceanography, 15, 336–347.

Millot, C. (1999), Circulation in the westernMediterranean Sea, J. Mar. Syst., 20, 423–442.

Millot, C., and A. Monaco (1984), Deep strongcurrents and sediment transport in the north-western Mediterranean Sea, Geo Mar. Lett., 4,13–17.

Moreno, A., J. Targarona, J. Henderiks,M. Canals,T. Freudenthal, and H. Meggers (2001), Orbi-tal forcing of dust supply to the North CanaryBasin over the last 250 kyrs, Quat. Sci. Rev.,20, 1327–1339.

Moreno, A., I. Cacho, M. Canals, M. A. Prins,M.-F. Sanchez-Goni, J. O. Grimalt, and G. J.Weltje (2002), Saharan dust transport andhigh-latitude glacial climatic variability: TheAlboran Sea record, Quat. Res., 58, 318–328.

Moreno, A., I. Cacho, M. Canals, J. O. Grimalt,and A. Sanchez-Vidal (2004), Millennial-scalevariability in the productivity signal from theAlboran Sea record, western MediterraneanSea, Palaeogeogr. Palaeoclimatol. Palaeoe-col., 211, 205–219.

Moreno, A., I. Cacho, M. Canals, J. O. Grimalt,M. F. Sanchez-Goni, N. Shackleton, and F. J.Sierro (2005), Links between marine andatmospheric processes oscillating on a millen-nial time-scale: A multi-proxy study of the last50000 yr from the Alboran Sea (western Med-iterranean Sea), Quat. Sci. Rev., 24, 1623–1636.

Nederbragt, A. J., and J. Thurow (2005), Geo-graphic coherence of millennial-scale climatecycles during the Holocene, Palaeogeogr.Palaeoclimatol. Palaeoecol., 221, 313–324.

O’Brien, S. R., P. A. Mayewski, L. D. Meeker,D. A. Meese, M. S. Twickler, and S. I. Whitlow(1995), Complexity of Holocene climate asreconstructed from a Greenland ice core,Science, 270, 1962–1964.

Oppo, D. W., J. F. McManus, and J. L. Cullen(2003), Deepwater variability in the Holoceneepoch, Nature, 422, 277–278.

Paillard, D., L. Labeyrie, and P. Yiou (1996),Macintosh program performs time-series ana-lysis, Eos Trans. AGU, 77(39), 379.

Palanques, A., N. H. Kenyon, B. Alonso, andA. Limonov (1995), Erosional and deposi-tional patterns in the Valencia channelmouth: An example of a modern channel-lobe transition zone, Mar. Geophys. Res.,17, 503–517.

Pinardi, N., and E. Masetti (2000), Variability ofthe large scale general circulation of the Med-iterranean Sea from observations and model-ling: A review, Palaeogeogr. Palaeoclimatol.Palaeoecol., 158, 153–173.

Prentice, I. C., S. P. Harrison, D. Jolly, and J. Guiot(1998), The climate and biomes of Europe at6000 yr BP: Comparison of model simulationsand pollen-based reconstructions, Quat. Sci.Rev., 17, 659–668.

Prospero, J. M. (1996), Saharan dust transportover the North Atlantic Ocean and Mediterra-nean: An overview, in The Impact of DesertDust Across the Mediterranean, edited byS. Guerzoni, and R. Chester pp. 133–151,Kluwer Acad., Dordrecht, Netherlands.

Reguera, I. (2004), Respuesta del MediterraneoOccidental a los cambios climaticos bruscosocurridos durante el ultimo ciclo glacial: Estu-dio de las asociaciones de foraminıferos, Ph.D.thesis, 231 pp., Univ. de Salamanca, Salamanca,Spain.

Rodo, X., E. Baert, and F. A. Comin (1997),Variations in seasonal rainfall in southern Eur-ope during the present century: Relationshipswith the North Atlantic Oscillation and theEl Nino-Southern Oscillation, Clim. Dyn.,13, 275–284.

Rohling, E. J., and H. Palike (2005), Centennial-scale climate cooling with a sudden cold eventaround 8,200 years ago, Nature, 434, 975–979.

Rohling, E. J., F. J. Jorissen, and H. C. de Stigter(1997), 200 year interruption of Holocene sa-propel formation in the Adriatic Sea, J. Micro-paleontol., 16, 97–108.

Rohling, E. J., A. Hayes, D. Rijk, D. Kroon, W. J.Zachariasse, and D. Eisma (1998), Abrupt coldspells in the northwest Mediterranean, Paleo-ceanography, 13, 316–322.

Rohling, E. J., P. A. Mayewski, R. H. Abu-Zied,J. S. L. Casford, and A. Hayes (2002), Holo-cene atmosphere-ocean interactions: Recordsfrom Greenland and the Aegean Sea, Clim.Dyn., 18, 587–593.

Rollinson, H. (1993), Using Geochemical Data:Evaluation, Presentation, Interpretation, 352pp., Longman Sci. and Technical, New York.

Ruhlemann, C., P. J. Muller, and R. Schneider(1999), Organic carbon and carbonate as paleo-productivity proxies: Examples from high andlow productivity areas of the tropical Atlan-tic, in Use of Proxies in Paleoceanography:Examples From the South Atlantic, edited byG. Fischer and G. Wefer pp. 1–31, Springer,New York.

Sanchez-Goni, M. F., I. Cacho, J. L. Turon,J. Guiot, F. J. Sierro, J.-P. Peypouquet, J. O.Grimalt, and N. J. Shackleton (2002), Syn-chroneity between marine and terrestrialresponses to millennial scale climatic variabil-ity during the last glacial period in the Medi-terranean region, Clim. Dyn., 19, 95–105.

Sbaffi, L., F. C. Wezel, G. Curzi, and U. Zoppi(2004), Millennial- to centennial-scale palaeo-climate variations during Termination I andthe Holocene in the central MediterraneanSea, Global Planet. Change, 40, 201–217.

Schulz, M., and A. Paul (2002), Holocene cli-mate variability on centennial-to-millennialtime scales: 1. Climate records from theNorth-Atlantic realm, in Climate Developmentand History of the North Atlantic Realm, edi-ted by G. Wefer et al., pp. 41–54, Springer,New York.

Shackleton, N. J., M. A. Hall, and E. Vincent(2000), Phase relationships between millen-nial-scale events 64,000–24,000 years ago,Paleoceanography, 15, 565–569.

Sierro, F. J., et al. (2005), Impact of iceberg melt-ing on Mediterranean thermohaline circulationduringHeinrich events,Paleoceanography, 20,PA2019, doi:10.1029/2004PA001051.

Skinner, L. C., and I. N. McCave (2003), Ana-lysis and modelling of gravity- and piston cor-ing based on soil mechanics, Mar. Geol., 199,181–204.

Skinner, L. C., and N. Shackleton (2003), Mil-lennial-scale variability of deep-water tem-perature and d18Odw indicating deep-watersource variations in the northeast Atlantic,0–34 cal. ka BP,Geochem. Geophys. Geosyst.,4(12), 1098, doi:10.1029/2003GC000585.

Steig, E. J. (1999), Mid-Holocene climatechange, Science, 286, 1485–1487.

Stow, D. A. V. (1982), Bottom currents and con-tourites in the North Atlantic, Bull. Inst. Geol.Bassin Aquitaine, 31, 151–166.

Stow, D. A. V., J. C. Faugeres, J. A. Howe, C. J.Pudsey, and A. R. Viana (2002), Bottom cur-

PA2209 FRIGOLA ET AL.: HOLOCENE CLIMATE VARIABILITY

15 of 16

PA2209

rents, contourites and deep-sea sedimentdrifts: Current state-of-the-art, in Deep-WaterContourite Systems: Modern Drifts and AncientSeries, Seismic and Sedimentary Characteristics,Geol. Soc. Mem., vol. 22, edited by D. A. V.Stow et al., pp. 7–20, Geol. Soc., London.

Stuiver, M., and P. J. Reimer (1993), Extended14C database and revised CALIB radiocarboncalibration program, Radiocarbon, 35, 215–230.

Telford, R. J., E. Heegaard, and H. J. B. Birks(2004), All age-depth models are wrong: Buthow badly?, Quat. Sci. Rev., 23, 1–5.

United Nations Environment Programme (2003)Riverine transport of water, sediments andpollutants to the Mediterranean Sea, MAPTech. Rep. Ser. 141, Athens.

Vanney, R., and D. Mougenot (1981), Laplate-forme continentale du Portugal et desprovinces adjancetes: Analyse geomorpholo-gique, Mem. Serv. Geol. Port., 28, 1–86.

Van Os, B. J. H., L. J. Lourens, F. J. Hilgen,G. J. de Lange, and L. Beaufort (1994), Theformation of Pliocene and carbonate cyclesin the Mediterranean: Diagenesis, dilutionand productivity, Paleoceanography, 9,601–617.

Velasco, J. P. B., J. Baraza, and M. Canals(1996), La depresion periferica y el lomo con-tourıtico de Menorca: Evidencias de la activi-

dad de corrientes de fondo al N del TaludBalear, Geogaceta, 20, 359–362.

Vernet, R., and H. Faure (2000), Isotopic chron-ology of the Sahara and the Sahel during thelate Pleistocene and the early and mid-Holo-cene (15,000–6,000 BP), Quat. Int., 68–71,385–387.

Voelker, A. H. L., S. M. Lebreiro, J. Schonfeld,I. Cacho, H. Erlenkeuser, and F. Abrantes(2006), Mediterranean outflow strengtheningduring Northern Hemisphere coolings: A saltsource for the glacial Atlantic?, Earth Planet.Sci. Lett., 245, 39–55.

Wehausen, R., andH.-J. Brumsack (1999), Cyclicvariations in the chemical composition of east-ern Mediterranean Pliocene sediments: A keyfor understanding sapropel formation, Mar.Geol., 153, 161–176.

Wehausen, R., and H.-J. Brumsack (2000), Che-mical cycles in Pliocene sapropel-bearing andsapropel-barren eastern Mediterranean sedi-ments, Palaeogeogr. Palaeoclimatol. Palaeoe-col., 158, 325–352.

Weldeab, S., W. Siebel, R. Wehausen, K.-C.Emeis, G. Schmiedl, and C. Hemleben(2003), Late Pleistocene sedimentation in thewestern Mediterranean Sea: Implications forproductivity changes and climatic conditionsin the catchment areas, Palaeogeogr. Palaeo-climatol. Palaeoecol., 190, 121–137.

Weltje, G. J., and M. Prins (2003), Muddled ormixed? Inferring palaeoclimate from size dis-tributions of deep-sea clastics, Sediment.Geol., 162, 39–62.

Zuo, Z., D. Eisma, and G. W. Berger (1991),Determination of sediment accumulation andmixing rates in the Gulf of Lions, Mediterra-nean Sea, Oceanol. Acta, 14, 253–262.

�������������������������I. Cacho, M. Canals, and J. Frigola, Con-

solidated Research Group Marine Geosciences,Department of Stratigraphy, Paleontology andMarine Geosciences, Faculty of Geology,University of Barcelona, Campus de Pedralbes,C/Martı i Franques s/n, E-08028 Barcelona,Spain. (miquelcanals@ub.edu)J. H. Curtis and D. A. Hodell, Department of

Geological Sciences, University of Florida,Gainesville, FL 32611-2120, USA.J. A. Flores and F. J. Sierro, Department of

Geology, University of Salamanca, Plaza de laMerced s/n, E-37008 Salamanca, Spain.J. O. Grimalt, Department of Environmental

Chemistry, ICER-CSIC, Jordi Girona 18,E-08034 Barcelona, Spain.A. Moreno, Pyrenean Institute of Ecology,

Spanish Research Scientific Council, Aptdo.202, E-50080 Zaragoza, Spain.

PA2209 FRIGOLA ET AL.: HOLOCENE CLIMATE VARIABILITY

16 of 16

PA2209